AI Accuracy & Reliability

AI is fast at drafting and unreliable at being right. This policy exists so that speed never reaches a learner as an unchecked claim.

What it covers

• AI-assisted slides, handouts, facilitator notes, exercises, assessments, examples and case studies.
• AI-generated or AI-supported live output during a session, including explanations, summaries, feedback and debriefs.
• Legal, regulatory, technical or statistical claims whose accuracy or currency could materially affect learning.

What we commit to

• Human subject-matter review is mandatory before materially AI-assisted learner-facing material is released.
• Review checks factual accuracy, alignment to the learning objectives, relevance to the audience, currency, source quality, unsupported claims and misleading simplifications.
• Legal and regulatory material is checked against primary or authoritative sources wherever available, with the verification date recorded where currency matters.
• Statistics, factual claims and citations are never accepted just because an AI tool supplied them. The reviewer checks the underlying source and removes, corrects or qualifies anything unsupported.
• Fast-moving AI content is refreshed at least monthly, and a significant tool, policy, regulatory or factual development triggers an out-of-cycle review.
• If an error surfaces during delivery, the trainer corrects it on the spot, stops further reliance on it and updates the source material. Where the error is material, corrected material is reissued.

What this means for you

A number, a legal reference or a claim on one of our slides has been checked by a person who knows the subject, and the check is recorded. If you think something is wrong, say so during the session; correcting it is part of the trainer's job, not an inconvenience.